Improving Multi-Robot Path Planning by Adaptive Artificial Potential Fields

The Artificial Potential Fields approach is amongst the widely used path planning methods in continuous environments. However, the implementation of it in multi-robot path planning encounters challenges such as the local-minima and an increase in traffic probability with the rise in the number of robots. The purpose of the proposed method is to improve multi-robot path planning in complex environments. A new adaptive potential function is introduced that reduces the probability of the robots entering an area at the same time and thus reducing the probability of traffic. Also, new potential functions have been proposed that lead to smoother paths with less traverse time when the robot encounters obstacles. In these functions, in addition to the position of robots and obstacles, heading of the robot and the position of the target are also considered. In order to evaluate this method, a distributed software architecture has been designed and implemented in the framework of the robot operating system. In this architecture, as robots move, new robots can join the operation or new tasks can be assigned to robots. Two series of real-time simulations are carried out in the Gazebo environment. The results show that the use of the proposed potential functions leads to a decrease in the convergence of the robots. In the simulation done for 2 robots, proposed method has resulted in a 35% reduction in the traversal time. While in case of 15 robots in the same map, a 50% reduction in the traversal time has been achieved.